560 Heritable Spiral Torsions. 
ting the long stalk of the terminal inflorescence, which 
is always present, I obtained the following series. 
PLANTS WITHOUT WITH 0~6 STRAIGHT INTER- 
TORSION NODES ABOVE THE TORSION 
Straight internodes 65432 1 
Individuals 900 2 3 1 2 40 200 148 
The curve constructed from these figures is obviously 
one with two peaks, and essentially the same as the cor- 
responding curve for fasciations seen in Fig. 118 (p. 521). 
Torsions with two, with one, and without any straight 
internodes above the torsions are far the commonest. 
Smaller torsions only occurred in nine cases among 1295, 
that is less than 1 % of all the individuals, or in about 
2% of the twisted individuals. I repeated the same ex- 
aminations in 1902; the intermediate forms were some- 
what more numerous and reached a proportion of about 
7% in 492 individuals. The form of the curve, was, 
however, not essentially modified. 
In order to determine the hereditary coefficients of 
the atavists I instituted an experiment in the eighth and 
ninth generation of my race. In July 1900, I had some 
highly twisted plants, some completely decussate ones, 
and some with ternary whorls only, on the same bed. 
Before the flowering period I reduced the inflorescences 
to the required number and afterwards insured pure fer- 
tilization in the following way. All the inflorescences 
were enclosed in parchment bags, and the bags were 
taken off the individuals falling into one group, for sev- 
eral hours one day out of three for each group. The 
humble bees flvmg about could thus fertilize only twisted 
* *> 
individuals on some days, on others only decussate, and 
on still others only ternary ones. This practice was 
continued until all the flowers were gone. The seeds 
